Overview
The Language Reader

A framework, and two indexes over it.

The books are not competing accounts of the same material. One states the theory; the other two cut the same language along different axes, so a phenomenon shows up once per series with a different question asked of it.

26 volumes493 sections49 catalogued defects, shown rather than hidden

The shelf volume numbers repeat across series by design

Three doors onto the same facts

By book — read a volume start to finish, as written. The corpus is good prose and stays readable linearly.By concept — land on a phenomenon and see all of its treatments at once: 40 of them, across 272 treatment records, every one cross-series.not built yetBy language — the dimension board, one row per dimension and one column per language, with the eleven unwritten languages as visibly empty columns.not built yet

Why the shape is not three columns

A concordance pass counted the links that already exist between the books. They run one way. El Árbol points at The Reference thirty-three times; The Reference points back zero times, and the one volume that maps the two books against each other omits the framework series entirely despite ten inbound links to it.

33El Árbol → The Reference
8The Reference → the framework
2El Árbol → the framework
0The Reference → El Árbol

Those missing reverse edges are the reason this is an app rather than a folder. Because the three run in parallel rather than in sequence, a volume is named by its series and its number together: three different volumes are each numbered 01, and they have nothing to do with one another.